Random code 29 on post

[ichee]

Hi. I have been having issues with post freezing on code 29 randomly when memory clocked higher than 2800Mhz. It occurs on cold boots, reboots, and wake from sleep. Tested with only memory overclocked to rule everything else out. Tried setting System Agent to between 0.784v - 1.14v with no effect, also vccio to between 0.95v - 1.14v.

The only thing that seems to affect the occurrence is Receiver DQ Pre-Emphasis and De-Emphasis. Tried between 0.7 and 1.1 with no effect, but 1.25 lasts one month before freeze occurs again.
Will setting higher cause any long term problems, or are there any other suggestions to help alleviate the post issues, besides lowering RAM speed?

RAM currently is OC to 3733Mhz at 17-17-17-36-1T, tRFC at 480, tREFI at 32767, all other second and third timings manually set to the auto defaults of 2800Mhz. vDIMM @ 1.29v, VCCSA @ 0.9v, VCCIO @ 1.0v.

All stable under Windows with 4hrs prime95 blend, 4 passes memtest86, 4hrs aida64 memory and cache only, 1500% HCI, 1.5hrs google stresstestapp, and 2 months of regular usage.

Thanks.

[Korth]

If the RAM isn't running too hot then you can bump the DRAM/vDIMM voltage up a little bit, 1.29V seems pretty low to me.
If the CPU isn't running too hot then you can bump the VCCSA a tiny tiny bit, 0.9V also seems a little on the low side to me.

You say the system runs "stable" but intermittently halts on code 29 during POST. Does this "random" halt always seem to occur in the same segment of the memory polling/count? If so, it could be a borderline/weak DIMM.

Have you tried reseating the DIMMs? And do any of them seem to be hotter than the others after code 29 halt?

Maybe not worth bumping any volts up at all, anyhow, if this halt cannot be reproduced on demand and only occurs on POST. Your description doesn't make it seem like you have to POST many times per day, lol. So, to be honest, I wouldn't even worry about it unless it begins to occur with annoying frequency or it actually crashes during runtime.

[ichee]

I’ve tried vDIMM @ 1.35v before without success, but not all three voltages increased at the same time. I’ve also tried setting vDIMM to 1.4v and eventual vDIMM to 1.29v, also without success.
Post seems to hang just before training occurs, and on rare occasion pressing reset after code 29 immediately results in a code 32. I’ve tried setting DRAM Training to both enabled and disabled. All seems to point to the IMC being unstable at boot, but is so intermittent it is difficult to tell. DIMM’s have been reseated several times, so I doubt that is a cause.
RAM is actively cooled by 2 120mm fans and temps are never more than 32-33C after hours of load.
Receiver DQ Pre-Emphasis and De-Emphasis seems to be the only thing that helps, but only at higher than 1.25.

POST stability is necessary for me because I automate a lot of tasks and depend on task scheduler to wake and run the system automatically, but if post fails, then I lost a day of work.

The x299 platform latency is high enough that an overclock is necessary to be certain that all scheduled tasks don’t miss their trigger, which has happened to me a few times at default speeds.

[Korth]

Sorry, don't know what else you could do, other than replace hardware.

Your DDR4 is a higher-rated bin than your actual configured speeds.
You don't want to tune them down because they impact your productivity.
Your volts/temps don't seem to be an issue.
Your i9-7900x CPU spec is rated for up 128GB quad-channel DDR4-2666, and your R6A mobo "officially" supports up to DDR4-4133.

But DDR4-3733 is still overspec and overclock, especially when populating 4 DIMMs, so no promises and no guarantees.

I think the limitation isn't on the mobo, although it could be (and if it is, it might be "fixable" with a future firmware/BIOS patch).
I think the limitation is the CPU, the iMC in your particular part just can't handle that much OC.

So I think that if automated startup/uptime/tasking is your priority then you simply have to lower your overclocked settings to something less spectacular but more reliable. Maybe remove some DIMMs, maybe tune down the frequency, maybe loosen up timings, depending on how these parameters affect your particular productivity performances metrics.

An alternative is to keep the machine powered on continuously so whatever conditions invoke POST code 29 cannot occur.
Another alternative is to link it to a second machine which remotely shutdown/restart your automated tasks and also monitor the status of the primary machine.
And, again, it doesn't hurt to identify if this issue can be isolated to a single DIMM. Or even to reseat all the DIMMs (and perhaps clean their contacts, if needed).

Somebody else (who actually has an ASUS X299, i9-X platform) might be able to offer you easy tweaks and better answers, lol.

[ichee]

Okay, then do you or anyone with experience with the Receiver DQ Pre-Emphasis and De-Emphasis know if setting it higher than 1.25, such as maybe 1.4 or higher, would cause any long term problems? Another user at http://www.overclock.net/forum/6-intel-motherboards/1635159-asus-rampage-vi-apex-x299-owners-77.html seemed to have a similar issue and was advised by Raja@ASUS to adjust the aformentioned settings with apparent success.

Also, can anyone with any RAM related POST issues on this platform who have been running for at least a few months suggest any voltage changes or timing adjustments that have stabilized POST?

Thanks.

[Quozzo]

DDR4 should be on 1.35v when overclocking/XMP but can go up to 1.45 or even 1.5v but I don't know if it's good for long term use, like years. You could use 1.45 (or 1.5v) to see if it's bootable at that voltage, if not it might be the Integrated Memory controller. Have you tried booting with 2 sticks and then the other 2?

[Raja]

ichee wrote:
Okay, then do you or anyone with experience with the Receiver DQ Pre-Emphasis and De-Emphasis know if setting it higher than 1.25, such as maybe 1.4 or higher, would cause any long term problems? Another user at http://www.overclock.net/forum/6-intel-motherboards/1635159-asus-rampage-vi-apex-x299-owners-77.html seemed to have a similar issue and was advised by Raja@ASUS to adjust the aformentioned settings with apparent success.

Also, can anyone with any RAM related POST issues on this platform who have been running for at least a few months suggest any voltage changes or timing adjustments that have stabilized POST?

Thanks.

What happens when you relax the memory timings? The XMP is likely configured for 2T, and while you are running the kit at a lower memory frequency, you're still pushing things with 17-17-17 primaries.

[ichee]

Timing changes seem to have no effect. I have tried running at their rated timings of 19-23-23-45-2T @ 1.35v with a lower frequency, but code 29 still occurs randomly. The only thing that seems to have an effect, other than reducing frequency to below 3000Mhz, is Receiver DQ Pre-Emphasis and De-Emphasis set to above 1.25. Could higher ram voltage have a negative effect on IMC stability during POST? I ask because all of the lower frequencies that do not have the issue are stable at 1.2v, even with tightened timings. Otherwise, is setting both Receiver DQ settings higher a viable option, or could it be potentially damaging?

Also, I noticed on an older UEFI version that when enabling XMP, Maximus Tweak was being set to mode 2, but there is no option for that on this board. Could there have been some change made between version 702 and 802 that affected compatibility negatively for this kit?

Quozzo wrote:
DDR4 should be on 1.35v when overclocking/XMP but can go up to 1.45 or even 1.5v but I don't know if it's good for long term use, like years. You could use 1.45 (or 1.5v) to see if it's bootable at that voltage, if not it might be the Integrated Memory controller. Have you tried booting with 2 sticks and then the other 2?

I wasn’t referring to ram voltage, but Receiver DQ.

[ichee]

Update:

A while back, I attempted to swap out my ram to see if the random code 29 was caused by that, and it still occurred.
About a week ago I got a new 9820x cpu and installed it, only to see another code 29 within a few days.
So at this point, the ram has been ruled out as well as the cpu’s imc, so unless there is a very specific configuration needed or a setting adjusted that has not, as of yet, been mentioned, then I’m starting to think the issue is with the board itself.

I have made every adjustment to all receivers and transmitters from 0.7 up 1.45, as well as vdimm, vccsa, and vccio running and boot voltages as high as 1.5v to no avail. Timings were loosened to 20-20-20-45-2T at 3200Mhz to rule out any timing issues. So unless someone can help give more insight as what may solve this issue, then I think an RMA is in order.